ADP Research argues AI’s productivity promise depends on how work changes

ADP Research argues AI’s productivity promise depends on how work changes

ADP Research published this article on September 22, 2026. It argues that AI’s main economic promise is not rapid job creation, but higher total factor productivity: making better use of workers and capital through improved processes, skills, technology and organization. The article contrasts labor productivity, output per hour worked, with total factor productivity, which captures changes not explained by labor, capital or output alone. ADP Research points to past productivity shifts to show why technology is only part of the story: digital investment and an experienced workforce helped boost productivity in the early 2000s, while economic and demographic headwinds weighed on it in the 1970s. For AI, the article says employment and wages can indicate whether workers are finding opportunities and sharing the gains. It also calls for a task-level view, linking pay to specific work activities, to make changes within jobs visible. ADP Research’s Canaries Dashboard, developed with Stanford Digital Economy Lab, tracks employment in occupations with varying AI exposure; its job-unbundling research uses payroll data to value specific activities. The article’s conclusion is that understanding AI’s effect on productivity requires examining employment, wages and tasks together. Whether AI-driven productivity gains will improve workers’ living standards remains unknown.

Publication: 36bf3993-5da7-ddb7-780e-b238e5b30a01

attribution — pass: The reading attributes the argument to ADP Research, the publisher of this single Main Street Macro commentary. The text names no individual author beyond a first-person 'My take' section, so publisher-level attribution is appropriate. Claims are framed as the article's arguments, not as established fact.

AI's big promise isn’t rapid job creation. Its real potential is to energize total factor productivity

summary — pass: The summary leads with the central thesis: AI's promise lies in total factor productivity, not job creation. It explains labor productivity and TFP in plain language, and the page also defines TFP and AI exposure. It covers the historical comparison and the proposed employment, wage and task-level indicators. The closing note about the unknown outcome is restrained and fits the source's monitoring stance. The question follows from the article's own framework.

But AI also demands that we add a third component, one that examines employment and wage data to link pay to distinct work activities.

date — pass: The reading gives September 22, 2026 at day precision, matching evidence.originalPublished. The 'week ahead' items about August employment data and the final September sentiment read are consistent with that date. The reading presents no later events or hindsight framing. The evidence is a publisher page, not an archive copy, so no archive disclosure is needed.

will provide data on August employment in occupations with varying levels of AI exposure

support — pass: Each claim matches its verbatim excerpt. The Canaries Dashboard and job-unbundling descriptions follow the text. The limitation correctly notes that the article does not show AI has already raised productivity. No unsupported benefits, costs or forecasts are added.

Our research on job unbundling uses ADP payroll data to assign value to specific work activities

media — pass: No media was selected, and no verified media candidates were supplied, so leaving media empty is appropriate.